OBJECTIVES: To apply to the treatment of experimental and clinical cancers the steadily accumulating concepts and methods of chronobiology; to develop for this purpose the necessary combination of biologic and computer methodology for assessing by inferential statistical means rhythmic markers (computer-analyzed reference variables). Thus: (a) To improve experimentally the therapeutic index of currently available carcinostatic therapy by: 1. timing administration according to rhythms known to occur under usual conditions in host tolerance and thus to specify a safer treatment time at which injury to the host is minimized and 2. exploiting experimentally information on the timing of rhythms in a sensitivity of certain tumors, thereby maximizing the desired curative effect. (b) To extend 1 and 2 above into clinical investigation at the outset with adriamycin and melphalan (each as monotherapy) and prednisone, ara-C and adriamycin or daunomycin in the treatment of certain leukemias. (c) To extend further experimentally the chronobiologic approach from single-agent treatment to drug combinations and combinations of drugs with radiation. (d) To achieve experimentally the best compromise between a.1. and a.2. for arriving at a favorable therapeutic ratio. (e) To further augment benefit from d. by developing new therapy involving the selective manipulation of rythms, at least in the host by meal scheduling and/or drugs (chronobiotics). (f) To attempt further augmentation of benefit by tumor rhythm manipulation, when such is found feasible. (g) To extend as many of the above steps as possible into clinical practice.